Gagan N Bangaragiri

Where I’ve worked

JUNE 2026 · NOW

AI Engineer · Genpact, Bangalore

Built AgenticDocExtractor, a 7-pass pipeline for digital and scanned invoice bundles. It uses PyMuPDF and Marker for digital PDFs, with Claude VLM and PaddleOCR for scanned PDFs.
Reduced LLM cost by 98%, from $2.65 to $0.05 per 10-page document, by running deterministic extraction before LLM fallback and caching identical model calls.
Implemented confidence-based routing with 95%+ auto-approval, an LLM judge for 60–95% confidence, and human review below 60%. Added versioned vendor schemas, PII-safe audit events, and a FastAPI and React review system backed by PostgreSQL.

PyMuPDFMarkerClaude VLMPaddleOCRFastAPIReactPostgreSQL

JANUARY · MAY 2026

Intern · Genpact, Bangalore

Developed internal GenAI systems for model evaluation and ETL migration with FastAPI, LangGraph, Azure AI Foundry, Databricks, PostgreSQL, Redis, and Docker.
Built reusable evaluation and migration services with automated validation, deployment checks, failure diagnosis, and support for enterprise data workflows.

FastAPILangGraphAzure AI FoundryDatabricksPostgreSQLRedisDocker

MAY · JULY 2025

Data Science Intern · Genpact, Bangalore

Developed two GenAI agents with LangChain and the Gemini API for risk model validation. The agents reduced manual validation time by 25% and improved issue logging accuracy.
Built an AI-powered VS Code extension with 10+ features for code explanation, refactoring, debugging, and other developer workflows.
Designed Figma prototypes and Streamlit applications for a Risk Model Validation UI and an Employee Document Chatbot.

LangChainGemini APIVS CodeFigmaStreamlit

A few things I’ve made (that I can show)

MASTER_CONTEXT.MDFOR DARREN

Master Context

Sprout is a fictional, local plant-watering reminder service. It reads configuration and plant records from files, calculates which plants need water, and exposes the result over HTTP.

Architecture Overview

Sprout is a small Python package with a file-backed read path.

sprout/ contains the HTTP server, reminder logic, and data migration.
settings.yaml supplies runtime settings. plants.json supplies sample plant records.
tests/ contains reminder and migration tests.

CHECKPOINT-CLEO.MDMON 5 OCT

Commit b26f83cContext

The watering interval setting is being renamed from reminder_days to watering_interval_days to make its purpose explicit.

Changes

settings.json: Replaces reminder_days with watering_interval_days, retaining the value of 7 days and port 8080.

Impact

due_plants now requires watering_interval_days; passing settings that contain only the old key raises a KeyError.

CATCHUP-AVA.MDFRI 9 OCT

While You Were Gone

You need to upgrade plant data, settings, dependencies, and client URLs before bringing an existing installation back up.

Critical Changes (Must-Read)

1. Migrate plant data before starting the service. Stop the service and run python -m sprout.migrate once.

2. Update reminder clients to GET /api/reminders.

3. Move settings to YAML and install PyYAML.

4. Rename the watering interval key.

5. Expect reminders on the scheduled day.

Checkpoint

Shipped

Project memory that stays with the code

FIG. 04 · Developer Tool / AI / Automation · 2026

Soteria

Building

Evidence before confidence

FIG. 02 · AI Reliability / Evaluation Infrastructure · 2026

Sisyphus

Shipped

A ranked-game chronicle for a Discord squad

FIG. 03 · Community Product / Automation / Data · 2026

What came before

August 2026

Vellore Institute of Technology

B.Tech, Computer Science and Engineering

Vellore, TN

July 2022

Greenwood High International School

X (ICSE) & XII (ISC)

Bangalore, KA

Tools I return to

Languages

PythonTypeScriptJavaScriptSQLJavaC++

AI & LLM

LangChainLangGraphLiteLLMAtomic AgentsAnthropic ClaudeGoogle Gemini APIAzure AI FoundryAWS Bedrock

ML & Vision

PaddleOCROpenCVYOLOv8SAM2spaCyScikit-learnMicrosoft PresidioHHEM-2.1-Open

Frameworks & Libraries

FastAPIReactStreamlitGradioSveltePydanticPyMuPDFOpenPyXLCeleryGitPython

Cloud & Infrastructure

GCP (Cloud Run, Cloud SQL, GCS, Cloud Build, Cloud Logging, Document AI)Azure (Document Intelligence, AI Foundry, OpenAI, Content Understanding)AWS BedrockDatabricksDockerPostgreSQLRedisSQLiteOpenBao

Developer Tools

GitGitHub ActionsAlembicPytestuvFigma

Spoken languages

English (Fluent)
Kannada (Native)
French (Conversational)
Hindi (Conversational)

Certifications

Supervised Machine Learning: Regression and Classification
DeepLearning.AI
Foundations of User Experience (UX) Design
Google Career Certificates
Start the UX Design Process: Empathise, Define, and Ideate
Google
Build Wireframes and Low-Fidelity Prototypes
Google